/
TODO.txt
33 lines (16 loc) · 874 Bytes
/
TODO.txt
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
TODO
- novel insight scoring (known bugs)
- when calculating results keep only 50% newest samples
- UI: disable all UI elements while optimizing
- renaissance:
* stacked RBM autoencoders do not work that well and require too much computing power (too slow)
=> implement variational autocoder (VAE) (see docs) to synthesize pictures?
- HMM based analysis and generation of N-step long probable stimulation sequences
(sample N>2 HMM sequences and only generate model and try
to learn 1..N long sequence models if probability of sequence is high enough)
- move from second long "ticks" to
subsecond ticks and use it everywhere to get subsecond accuracy
- get emotiv insight and add support for it
- create theora+ogg video+audio files when measuring programs
(currently does not
- improve sound synthesis modules (FM synthesis)